image interpretation theory.pdf
TRANSCRIPT
-
7/29/2019 Image interpretation theory.pdf
1/9
ITI 230ITI 230 -- HCIHCI04:547:230(01) (Index: 10636)
Lecture #11Lecture #11
Information DesignInformation Design
Dr. Jacek Gwizdkahttp://www.scils.rutgers.edu/~jacekg/teaching/
ITI 230 HCI Jacek Gwizdka 2
Iterative Design ProcessIterative Design Process
1. Identify user needs and establish requirements2. Develop alternative designs3. Build interactive prototypes each time more refined
to evaluate them to communicate them to other people (managers, users)
4. Evaluate the prototypes1. with Experts2. with Users3. involving no people
5. Repeat the process (go to 2)
ITI 230 HCI Jacek Gwizdka 3
Requirements
Future TrendsEnd!
Evaluation w.experts
Lo-Fi & Hi-FiPrototyping
ConceptualDesign
Week 5
Week 7
InfoDesign
Week 11Week 4
Week 8
User Needs & Requirements
Evaluationw. users
ColorDesign
Week 12
Info Architecture
Week 12
Week 3
Understanding users Human Cognition
Week 14 & 15
Evaluationw/o people
Design
EvaluatePrototype /Implement
Iterative Design ProcessIterative Design Process
Week 5
Week 6&7
ITI 230 HCI Jacek Gwizdka 4
TodayToday ss OverviewOverview
Information Design What is it? Information Design to support :
information perceptionBertins visual attributesGestalt Principles
information interpretation making sense of information in the current context
ITI 230 HCI Jacek Gwizdka 5
Information DesignInformation Design
Definition: During Information Design objects andactions possible in system are created and arrangedto facilitate perception and understanding
(Visual) Information Design includes: design of screens, web pages, menus, dialog boxes, icons, etc.
ITI 230 HCI Jacek Gwizdka 6
NormanNorman s Action Cycles Action Cycle
Human action has two primary aspects Execution : doing something Evaluation : comparison of what happened to what was
desired
-
7/29/2019 Image interpretation theory.pdf
2/9
ITI 230 HCI Jacek Gwizdka 7
NormanNorman s Action Cycles Action Cycle (review)(review)
Norman, 1986ITI 230 HCI Jacek Gwizdka 8
NormanNorman s Action Cycles Action Cycle
Evaluation has three stages:1. Perception
2. Interpretation3. Making sense of information (evaluation of interpretations)
Goals of Information Design, support:1. perception,2. interpretation,3. sense-making (understanding)
ITI 230 HCI Jacek Gwizdka 9
1.1. Perceiving InformationPerceiving Information
Pixels, colors or tones are designers palette
These elementary graphical elements are created andarranged to convey higher-level structures
Goal: to guide users to see structure in informationdisplay
ITI 230 HCI Jacek Gwizdka 10
PerceptionPerception -- exampleexample
Guiding users to see structure in information display
(Tufte1983, 49: Source: Ho, Powell, and liley. Journal of Physical and Chemicalreference Data 3 , Supplement, 1, 1974, American Institute of Physics and NIST )
ITI 230 HCI Jacek Gwizdka 11
Visual Design to Aid Perception Visual Design to Aid Perception
Elements of visual alphabet Visual variables & perception stylesBertin
Organization and visual structure Principles of perceptual organizationGestalt principles
ITI 230 HCI Jacek Gwizdka 12
BertinBertin s Taxonomys Taxonomy
Bertins taxonomy includes (Bertin 1981; 1983) seven visual (retinal) variables four perception styles
Visual variables : shape, orientation, colour,texture, value, size & position All are perceived quickly and effortlessly across visual f ield
Perception styles : associative, selective, ordered,& quantitative
-
7/29/2019 Image interpretation theory.pdf
3/9
ITI 230 HCI Jacek Gwizdka 13
BertinBertin s Visual Variabless Visual Variables
Illustration of
visual variables(position not shown)
ITI 230 HCI Jacek Gwizdka 14
BertinBertin s Perception Styless Perception Styles Associative
element can be immediately isolated as belonging to the same category visual variable is associative if it does not affect visibility of other
dimensions of visual objects (it can be ignored with no effort)e.g. color is visible regardless of objects orientation
Selective element can be grouped into a category differenced by this variable variable is selective if grouping is immediate and effortless
Ordered allows each element to be grouped into an order of scale variable is ordered if ranking can be read without referring to key
Quantitative each visual element can be compared to be greater or smaller than
another element variable is quantitative if relative magnitude can be judged without
referring to key or index
ITI 230 HCI Jacek Gwizdka 15
Visual Variables vs. Perception Styles Visual Variables vs. Perception Styles Visual variables vary in their aptness for the four
perception styles
X
X X
X X
X
XX
X
XX
ITI 230 HCI Jacek Gwizdka 16
Visual Variables vs. Perception Styles Visual Variables vs. Perception Styles-- ExamplesExamples
Only shape does not support selective perception
Only position and size are quantitative
Colour is neither ordered not quantitative
1x2x
31
2
3 445
6
6 6
68
987 7
12
0
02
0
331
2
3 44
5
6
66
68
987 7
12
0
02
0
3
ITI 230 HCI Jacek Gwizdka 17
CognitionCognition PerceptionPerception
A picture is worth a thousand words
Make information perceptually explicit information processing load is shifted from the users
cognitive system to the perceptual system (Larkin & Simon, 1987)
graphical representation can be more computationallyefficient than textual representation
ITI 230 HCI Jacek Gwizdka 18
Offloading Cognition exampleOffloading Cognition exampleWebTaskMail email interface (Gwizdka 2004)
Graphical representation requires less effort Information equivalence Computational (mental effort) equivalence (Herb Simon)
UI- Visual UI- Text
-
7/29/2019 Image interpretation theory.pdf
4/9
ITI 230 HCI Jacek Gwizdka 19
Task vs. Visual RepresentationTask vs. Visual Representation Date information was found faster in UI-Visual
Use graphical representation it requires less effort Information equivalence Computational (mental effort) equivalence(Herb Simon)
2nd session: Effect of UI * Task on t ime
00:00
00:10
00:20
00:30
00:40
00:50
Head er Date M ixe d Ta s k
time sec
UI TextUI Visual
ITI 230 HCI Jacek Gwizdka 20
Which representation is best?Which representation is best?
It depends on the user task!
What is the precisevalue of sysperformance?
How does the current performancecompare to its peak? How has performance
changed over time?
Based on material prepared by RavinBalakrish nan for similarcourses at UofT. Used with the permission of the original author.
System performance: processor and file system
ITI 230 HCI Jacek Gwizdka 21
Gestalt PrinciplesGestalt Principles
German psychologists in 1920s introduced the concept of figure and ground in perception
Gestalt principles of perceptual organization describeconfigural properties of visual information how individual bits of information are grouped together what elements will be seen as a coherent figure , and what elements will appear as background
ITI 230 HCI Jacek Gwizdka 22
Gestalt PrinciplesGestalt Principles
A page of paragraphs, a grid of spreadsheet cells, a left-justified list of selections or parameters
Continuity: there is a tendency to groupelements into continuous contours orrepeating patterns
Window manipulation controls (e.g.,scroll bar, selection handles)
Symmetry: there is a tendency to seesymmetric elements as part of the samefigures
Icons on a workstation screen, pop-upmenu on top of a document display
Area: there is a tendency to groupelements in a way that creates thesmallest possible figure
Overlapping windows, menus, dialogboxes and other user interface controls
Closure: there is a tendency to organizeelements into complete, closed figures
Toolbar icons (proximity operates hereas well), data visualization
Similarity: elements that share visualcharacteristics (shape, color, etc.) tend tobe seen as a group
Words on a menu bar, columns in atabular display, text in a paragraph
Proximity: elements near each othertend to be seen as a group
User Interface ExamplesGestalt Principle
Rosson & Carroll, 2002
ITI 230 HCI Jacek Gwizdka 23
Gestalt PrinciplesGestalt Principles Examples 1Examples 1
Proximity
Similarity
ITI 230 HCI Jacek Gwizdka 24
Gestalt PrinciplesGestalt Principles Examples 2Examples 2
Closure
Area (smallness)
Symmetry
-
7/29/2019 Image interpretation theory.pdf
5/9
ITI 230 HCI Jacek Gwizdka 25
Gestalt PrinciplesGestalt Principles Examples 3Examples 3
Continuity
ITI 230 HCI Jacek Gwizdka 26
Principles @ Work Principles @ Work UIUI
Dialog box used to set and preview page borders in Microsoft Word (example from Rosson& Carroll, 2002)
ITI 230 HCI Jacek Gwizdka 27
Principles @ Work Principles @ Work UIUI
User should perceive immediate sense of organization and seeindividual control elements (Rosson & Carroll, 2002)
TRADEOFF:Displaying all active task entities enhances understanding and feelings of control,BUT each display element adds to the complexity of the user interface design.
SOLUTION: Decompose and link related informationFor example: Use tabs, Link dialogs (one dialog opens another)
ITI 230 HCI Jacek Gwizdka 28
Principles @ Work Principles @ Work UIUI
Proximity and Similarity
ITI 230 HCI Jacek Gwizdka 29
Principles @ Work Principles @ Work UIUI Use contrasting visual features to create different groups
TRADEOFF:Perceptual distinctions highlight and group user interface elements, BUT the moredistinctions a display uses, the less noticeable and useful any individual cue will be.
SOLUTION: Limit the number of perceptual distinctions used at any one time.(also related to the previous solutiondecompose)
ITI 230 HCI Jacek Gwizdka 30
Principles @ Work Principles @ Work UIUI (Area) White Space
(White space)
-
7/29/2019 Image interpretation theory.pdf
6/9
ITI 230 HCI Jacek Gwizdka 31
Principles @ Work Principles @ Work UIUI Borders and bounding boxes provide further structural cues
3
2
bad example from Jeff Johnson 2000, GUI Bloopers ITI 230 HCI Jacek Gwizdka 32
Principles @ Work Principles @ Work UIUI Redundant coding - use of multiple perceptual cues Redundancy increases the speed and accuracy of perceptual
processing
ITI 230 HCI Jacek Gwizdka 33
Principles @ Work Principles @ Work UIUI
Use grid-based design
Benefits : all elements are aligned and symmetrically balanced (vertical axis) regular and predictable information structure across all screens (pages)
Aaron Marcus and Associates for Kaiser Permanente
ITI 230 HCI Jacek Gwizdka 34
Fonts and FormattingFonts and Formatting Dont use too many fonts, in terms of typeface and in size.
The general practice: limit to 3 different type settings . Variable width fonts such as Times New Roman are easier to read thanfixed width fonts such as courier. STUDIES SHOW THAT TEXTIN ALL CAPS SLOWS READING BY 12% SO YOU SHOULD MIX CASE.
Reading on-screen: sans-serif fonts are easier to read (e.g. Verdana ,Tahoma course slides, Arial )
Specially design serif fonts are good too (e.g. Georgia course website)
The same goes for justified text, so that issomething of which to beaware. Al so short
justified lines are to beavoided, but that shouldbe common sense.
Based on material prepared by Elaine Toms for similar courses atUofT. Used with the permission of the original author.
ITI 230 HCI Jacek Gwizdka 35
2.2. Interpreting InformationInterpreting Information
Perception users see lines, rectangles, text strings Perceiving higher-level structure in information display (groups
of objects) enables interpretation Interpretation associate meaning with display elements
ITI 230 HCI Jacek Gwizdka 36
Facilitating InterpretationFacilitating Interpretation
Using familiar words and visual language Refining visual language
Abstract images vs. realistic images
Employing affordances
-
7/29/2019 Image interpretation theory.pdf
7/9
ITI 230 HCI Jacek Gwizdka 37
FamiliarityFamiliarity
Familiar words and visual language aids interpretation familiar terms are usually less distinctive and precise
e.g. enter submit this data or come into this space
TRADEOFF: A familiar vocabulary eases interpretation, BUT familiar terms are usually lessdistinctive and precise, and what is familiar for one person may not be familiar foranother.
SOLUTION: Carefully select words and imagery. Be aware of individual and culturaldifferences.
ITI 230 HCI Jacek Gwizdka 38
Abstract vs. Realistic Images Abstract vs. Realistic Images
Realistic images can be recognized accurately, but recognitionis more complex (and interpretation may be slower)
Recognizing abstract images requires (some) learning, butrecognition process is simpler and faster
TRADEOFF:People's recognition memory for realistic images is accurate and long-lasting, BUTrealistic images are often complex, and suggest instances rather than concepts.
SOLUTION: (User task and technology dependent.) In most cases, abstract images(e.g. icons) will improve visual design. Interpretation of abstract images isenhanced when a set of icons is refined together.
ITI 230 HCI Jacek Gwizdka 39
Recognizing AffordancesRecognizing Affordances
Affordances must be visible to aid interpretation More competition for screen space and more complexity
TRADEOFF:Dynamic hiding of user interface controls that are not in use reduces complexityand simplifies the display, BUT also conceals the controls' affordances from usersand requires that users know that controls are available and how to activate them.
SOLUTION: Use dynamic affordances.
ITI 230 HCI Jacek Gwizdka 40
3.3. Making Sense of InformationMaking Sense of Information
Relating interpreted information to current task
If users cannot make sense of info they may trydifferent interpretations
ITI 230 HCI Jacek Gwizdka 41
Making SenseMaking Sense -- ExampleExample
Requirements to make sense of Minards visualization knowledge of maps in general (Eastern Europe and rivers in particular) knowledge of the Russian campaign (e.g., that many men were lost) only then appropriate understanding is possible:
bands = travel paths plotted on abstracted mapsize of paths = size of marching army
Minard's visualization of Napoleon's Russian campaign
ITI 230 HCI Jacek Gwizdka 42
Information IntegrationInformation Integration
Making sense of information requires informationintegration
User connects new information to: previously displayed information
e.g. sequence of screens or web pages
current goals and task users domain expertise users knowledge about world
-
7/29/2019 Image interpretation theory.pdf
8/9
ITI 230 HCI Jacek Gwizdka 43
Facilitating Making SenseFacilitating Making Sense
Consistency Visual metaphors Information models Dynamic displays
ITI 230 HCI Jacek Gwizdka 44
Consistency Consistency
Consistent use of terminology and symbols e.g. use Move backward with Move forward, not Reverse
Consistent use of visual vocabulary shapes, fonts, colors, character of icons
Consistent layout e.g. OK/Cancel buttons are typically in bottom-right corner but some dialog boxes use other buttons
Scope of consistency internal (within this program) external consistency (other applications)
TRADEOFF:Consistent use of shapes, sizes, colors, and position sets up expectations that helpusers make sense of information, BUT in some cases consistency may actuallyimpair user performance or satisfaction.
SOLUTION: Dont be afraid to break consistency when needed
ITI 230 HCI Jacek Gwizdka 45
MetaphorsMetaphors
Consistency with external world Metaphors should balance
consistency and familiarity withinconsistency and innovation
Technology should enhancephysical world metaphorsshould not be taken too literally
TRADEOFF: Visual metaphors leverage real-world knowledge of objects, BUT metaphors thatare very literal may introduce information that is irrelevant or misleading to a task.
SOLUTION: Break metaphors when necessary
ITI 230 HCI Jacek Gwizdka 46
Information ModelInformation Model Set of concepts, relationships, representations developed to
help make sense of large data sets
Common Info Model - Hierarchy used to organize navigation through complex info spaces common use of hierarchies: library classification, menus, web sites appropriate decomposition (categorization) and use of familiar concepts
at each level of hierarchy are criticalexamples of decomposition: Task-centric, Chronological, Spatial,Organizational, Alphabetical, Frequency of use
breadth & depth: number of choices at each level and number of levels information architecture
Information visualization visual features are used to code data attributes
ITI 230 HCI Jacek Gwizdka 47
Dynamic DisplaysDynamic Displays
Computer displays can dynamically change content Restructuring, resizing, dynamic filtering, or
animating can help to make sense of information
Dynamic displays allow for exploration
ITI 230 HCI Jacek Gwizdka 48
Dynamic DisplaysDynamic Displays FishFish -- eye Viewseye Views Fish-eye view (Furnas 1986)
overview with local detail (dynamic)
Bederson, B.B. (May 2000)University of Maryland
Inxight inc. table lens
-
7/29/2019 Image interpretation theory.pdf
9/9
ITI 230 HCI Jacek Gwizdka 49
Dynamic DisplaysDynamic Displays HierarchiesHierarchies
Cone Trees and Hyperbolic Lens
Robertson / Mackinlay / Card Cone Trees: Animated 3D Visualizations of HierarchicalInformation. ProcACM CHI'91
A hyperbolic browser (Inxight Tree Studio) used tovisualize Web pages from Barnes & Noble
ITI 230 HCI Jacek Gwizdka 50
ReducingReducing Amount Of Info Amount Of Info
Infinite zoomZoomable UI (ZUI)
Semantic filtering(not shown)
Pad++: A Zoomable GraphicalSketchpad for Exploring AlternateInterface Physics Bedersonet alJournal of Visual Languages and Computing 7, 1996
ITI 230 HCI Jacek Gwizdka 51
SummarySummary Information Design is about creating and arranging
display objects to facilitate perception, interpretationand understanding
Norman's action cycle is a useful framework foranalysis of information design
Bertins taxonomy and Gestalt principles of perceptual organization offer useful guidance inselecting and arranging elements of visual display
The overall information model, how it maps to users'mental model of the task, and its consistencydetermine how well users will be able to integrate
and make sense of their actions as they progressthrough a task ITI 230 HCI Jacek Gwizdka 52
Next Week Next Week
Color in Design